An Investigation of Atmospheric Stability and Its Impact on Scatterometer Winds Across the Gulf Stream

Jim EdsonUniversity of Connecticut

Doug Vandemark & Amanda PlaggeUniversity of New Hampshire

CLIMODE Deployments and Cruises

• November 2005: Mooring & Profiler Deployment Cruise

• January 18-30, 2006: Pilot Experiment, ASIS/FILIS Deployment

• October 2006: Mooring Turnaround Cruise

• February-March 2007: 6-week Main Experiment, ASIS/FILIS Deployments, Microstructure, Surveys.

• November 2007: Mooring Recovery Cruise

CLIMODE Deployments and Cruises

• November 2005: Mooring & Profiler Deployment Cruise

• January 18-30, 2006: Pilot Experiment, ASIS/FILIS Deployment

• October 2006: Mooring Turnaround Cruise

• February-March 2007: 6-week Main Experiment, ASIS/FILIS Deployments, Microstructure, Surveys.

• February 2007: Mooring Recovery Cruise

Cold A

ir Outbreaks

L

The Gulf Stream

• Cold air outbreaks drive extremely active convection over the region.

22oC

14oC

• The net winter heat loss in this region is 400 W/m2.

CLIMODE Platforms

Sensor Packages• R/V Atlantis and Knorr

– 2-3 DCFS (Sonic/MotionPak/Licor)

– IR and Solar Radiometers

– IR SST

– RH/T/P Sensors

– ShipSystem (Precip, Tsea, Salinity, ADCP)

• ASIS– DCFS (Sonic/MotionPak/Licor)

– IR and Solar Radiometers

– RH/T/P Sensors

– 6 Wave Wires

– Subsurface (Tsea, Salinity, ADCP, Nortek)• Discus

– Low Power DCFS (Sonic/MotionPakIII)

– Redundant IR and Solar Radiometers

– Redundant U/RH/T/P Sensors (ASIMET)

– Subsurface (T/S, Nortek, VACM)

Moving Platform vs. Fixed Tower

Uncorrected

~530 m

Moving Platform vs. Fixed Tower

Corrected

~530 m

Bulk Aerodynamic Method

Latent Heat Flux: Lv <wq> LvCE UQ

Sensible Heat Flux: cp <w> cpCH U

Momentum Flux: -<uw> - CD U2

Direct Covariance Bulk Aerodynamic

Drag Coefficient Formulas

• Semi-empirical

Large & Pond (1981)

• “Empirical”

Wind Speed Dependent

TOGA-COARE 4.0

AtmosphericStability

Surface Roughness

2

20 )/()/ln()/,/(

LzzzU

uwLzzzC

moD

2

2

0 )/ln()/(

NoDN

U

uw

zzzzC

)(10 103

NDN UC-1

10 ms 1142.1 NU-1

1010 ms 2511065.049.0 NN UU

Drag Coefficient Formulas

• Semi-empirical

TOGA-COARE 4.0

AtmosphericStability

Surface Roughness

2

20 )/()/ln()/,/(

LzzzU

uwLzzzC

moD

2

2

0 )/ln()/(

NoDN

U

uw

zzzzC

• COARE parameterizes the roughness length as:

Charnock Parameterg

uU

uzo

2*

10*

)(

g

uU

uzo

2*

10*

)(

MBL/CBLAST/CLIMODE Drag Coefficients

Wave Age Dependent Drag

007.00036.0/ 10* Np Ucu

Wave Age Dependent Drag

Bp ucA )/( * 007.00036.0/ 10* Ucu pplus equals ECMWF

Flux Time Series

Summary• A wind speed dependent drag coefficient give good results over a

wind range of sea-states/wave-ages.– This requires a wind speed dependent Charnock variable– Numerous investigations have shown that the Charnock variable is

dependent on wave-age.– However, these findings can be reconciled since observed wave ages over

the coastal and open ocean are clearly associated with wind ranges.

QuikSCAT Wind Speeds

QuikSCAT vs. Buoy Wind Direction

QuikSCAT vs. Buoy Wind Speeds

Atmospheric Forcing

Sikora et al. (1995)

PO.DAAC

Stability Effects Near SST Fronts• Boundary Layer Adjustment

– Baroclinic adjustment to horizontal temperature gradients.

– Acceleration/deceleration of surface winds.

• Surface Layer Adjustment– QuikSCAT measures surface roughness/stress

– Surface stress is proportional to neutral winds, UN

• UN < U in unstable conditions

• UN > U in stable conditions

• Mesoscale Adjustment to SST fronts– Combination of both?

QuikSCAT vs. Buoy Wind Speeds

U10

U10

“Surface Layer Adjustment”

U(z) = u*/κ[ln(z/zo) – ψm(z/L)]

QuikSCAT vs. Buoy Wind Speeds

U10 U10N

U10 U10N

“Surface Layer Adjustment”

U(z) = u*/κ[ln(z/zo) – ψm(z/L)] UN(z) = u

*/κ[ln(z/zo)]

QuikSCAT vs. Buoy Wind Speeds

U10 U10N

U10 U10N

“Surface Layer Adjustment”

U(z) = u*/κ[ln(z/zo) – ψm(z/L)] UN(z) = u

*/κ[ln(z/zo)]

QuikSCAT vs. Buoy Wind Speeds“Surface Layer Adjustment”

U(z) = u*/κ[ln(z/zo) – ψm(z/L)] UN(z) = u

*/κ[ln(z/zo)]

Smaller than measured

Larger than measured

QuikSCAT vs. Buoy Wind Speeds“Surface Layer Adjustment”

U(z) = u*/κ[ln(z/zo) – ψm(z/L)] UN(z) = u

*/κ[ln(z/zo)]

Baroclinicity?

QuikSCAT vs. Buoy Wind Speeds“Boundary Layer (Baroclinic) Adjustment”

Coupling Coefficients

O’Neill et al. (submitted)

30 Day Perturbations

Courtesy of JHU/APL

QuikSCAT vs. Buoy Wind Speeds

O’Neill et al. (submitted)

“Boundary Layer (Baroclinic) Adjustment”

QuikSCAT vs. Buoy Wind Speeds

O’Neill et al. (submitted)

“Boundary Layer (Baroclinic) Adjustment”

QuikSCAT vs. Buoy Wind Speeds

No obvious trend in perturbations when computed versus sea-air virtual temperature difference.

“Boundary Layer (Baroclinic) Adjustment”

QuikSCAT vs. Buoy Wind Speeds

However, it becomes more obvious when you only look at cold/cool air advection.

JHU/APL

“Boundary Layer (Baroclinic) Adjustment”

QuikSCAT vs. Buoy Wind Speeds

JHU/APL

“Boundary Layer (Baroclinic) Adjustment”

However, it becomes more obvious when you only look at cold/cool air advection.

Summary• A wind speed dependent drag coefficient give good results over a

wind range of sea-states/wave-ages.– This requires a wind speed dependent Charnock variable– Numerous investigations have shown that the Charnock variable is

dependent on wave-age.– However, these findings can be reconciled since observed wave ages over

the coastal and open ocean are clearly associated with wind ranges.

• Some of the variability in the QuikSCAT winds is due to adjustment of the neutral wind to changes in stratification and not changes in the actual wind speeds.– This variability obeys MO-Similarity in the mean.– This effect enhances the gradient in neutral winds but not actual.– Significant variability in the QuikSCAT winds is not explained by this effect

• The one-buoy approximation of the coupling coefficients is in reasonably good agreement with previous studies.– This includes the neutral wind, measured wind, and directly measured stress.– The physical processes responsible for this correlation is …

• Compare stress!

Thanks to NSF and NASA for supporting this research.

QuikSCAT vs. Buoy Wind Speeds

U10 U10N

U10 U10N

“Surface Layer Adjustment”

U(z) = u*/κ[ln(z/zo) – ψm(z/L)] UN(z) = u

*/κ[ln(z/zo)]